Umbrella review

An umbrella review is a review of reviews. It collects every systematic review and meta-analysis on a broad topic and compares them side by side, instead of going back to the individual studies. John Ioannidis set out the format in 2009 as a way to see a whole research field at once. The unit of analysis is the meta-analysis, not the trial. For each association the authors re-extract the pooled effect, the number of cases, the heterogeneity statistic I², the 95% prediction interval, and tests for small-study effects and excess significance. Those feed a credibility grading. The top tier, usually called convincing evidence, requires more than 1,000 cases, a p-value under 10⁻⁶, I² below 50%, a prediction interval that excludes the null, and no sign of small-study bias. Most associations fail it. A 2014 umbrella review of vitamin D worked through 107 systematic reviews and found no outcome with highly convincing evidence. The limits are inherited: weak primary studies cannot be repaired afterwards, and overlapping reviews count the same trials more than once. For longevity claims about supplements or single foods, this format shows fastest how thin the underlying evidence usually is.

Last reviewed: August 25, 2026

This definition is educational and is not medical advice, a diagnosis, or treatment. Talk to a doctor about any health decisions. Read our full medical disclaimer